Get started with ModelMart

One API key, one endpoint, every model.

1. Create an API key

Create an account in the dashboard and issue an API key. Keys look like mm_.... Keep the key in an environment variable, never in source code or a browser bundle.

2. Make a request

ModelMart uses the OpenAI-compatible chat completions format. The main endpoint is https://api.modelmart.ai/v1.

curl https://api.modelmart.ai/v1/chat/completions \
  -H "Authorization: Bearer mm_..." \
  -H "Content-Type: application/json" \
  -d '{"model":"claude-sonnet-5","messages":[{"role":"user","content":"Hello"}]}'

3. Choose a model

Select a model with the model field. Example first-party and Verified models include claude-sonnet-5, gpt-5.6-sol, gemini-3.6-flash, and deepseek/deepseek-v4-pro.

Community models such as Qwen/Qwen3.7-Max run on an endpoint operated by a seller. That seller may see your prompts and responses. Use Community only after accepting the warning and enabling allow_community for the API key.

4. Read usage

Responses follow the OpenAI shape, including choices and usage when the provider supplies usage data. ModelMart uses token counts to settle the request cost. There is no shared response cache.

Next steps

Read authentication, models, and errors before shipping to production.